Our country is a big agricultural country, and the healthy development of agriculture is relating to the roots of the state. In recent years, with the improvement of living standard, customers demand for the quality and safety of agricultural products continuously. Modem agriculture is the organic combination of physical technologies and agricultural production, which is an environment control type agriculture that controls the living environments and growing development of animals and plants through such physical factors having biological effects like electricity, sound, light, magnetism, heat, nucleus, or the like, promotes the conventional agriculture to gradually get rid of the dependence on such chemicals like chemical pesticides, chemical fertilizers and antibiotics, and finally obtains non-toxic agricultural products with good quality and high volume. Therefore, the principal job for predicting injurious insects is to recognize the injurious insects and accurately obtain the quantity information.
Conventional injurious insect recognition and counting are mainly subjected to an artificial recognition method, a field survey method, a trapping method, or the like. The artificial recognition and counting severely has such defects like low recognition rate, poor counting accuracy, high field task labor intensity and being not in real time due to such factors like complicated and instable field environments, and this method cannot satisfy the monitoring requirements on a condition that the current injurious insects in the fields are severe. The field survey method consumes time and effort, and has more procedures in data investigation, recording and reporting, has high workload for monitoring personnel, is largely affected by subjective factors, and has poor data application timeliness, which affects the accurate prediction of the injurious insects, and cannot satisfy the actual production requirements. In order to solve the problems of time and effort consuming for recognizing the injurious insects and obtaining the quantity data, scientists are continuing to explore new technologies for automatically recognizing and counting injurious insects.
With the development of computer technology, microelectronic technology or the like, a great progress has been made on technologies for automatically recognizing and counting injurious insects. At present, the technologies for automatically recognizing and counting injurious insects mainly include sound signal, image technology, infrared sensor, or the like. The development of these technologies improve the efficiency for automatically recognizing and counting injurious insects, promotes the implementation of accurate agriculture, reduces the losses caused by the injurious insects, reduces the environment pollution, and improves the be comprehensive control level of the injurious insects. However, although the infrared counting method has a quick speed and can record the quantity of the insects trapped, it is weaker in recognizing the types of the injurious insects, and is easily disturbed by other articles fell into. Because the injurious insects trapped in fields are different in gestures and are different to master samples, the image processing method which is subjected to classifier training on the basis of the morphological specificity of a standard gesture, is easy to weaken the classifier generalization ability, and is not beneficial for recognizing the injurious insects and counting the quantity thereof.
The above solutions for counting injurious insects are either complicated or counts inaccurately.